[LINK] RFI: Boomerang Traffic

[Frank O'Connor]

Mmmm,

In theory the any given data path should correspond with the lowest hop-count, but this rarely applies in today's modern and private networks.

I used to note that my traffic to the US and Europe got routed through West Australia and the Indo-Chinese/Japanese routers when I was with iiNet, Telstra favours the Pacific route and other providers presumably try to route their international traffic through whatever pipes they own/lease/control.

In country, various peering and other alliances and arrangements can also result in seemingly illogical routes and higher hop counts ... with monotonous regularity. I've had e-mail (probably the easiest protocol to route check) from here in Victoria routed through Sydney, and mail-servers seem to change change network locations with monotonous regularity. It's strange that a mail item to a bloke down at Rosebud from here in Rye (6 or 7 miles away geographically) can be routed through Sydney ... but it has happened. Network and geographic locations only rarely correspond. And mail servers can be sited anywhere.

I'm guessing that business exigencies and economies, ISP contracts and agreements with providers 'down the line', peering arrangements, and a whole heap of other commercial realities get in the way of how TCP/IP and its various application protocols are supposed to work .... which is probably not surprising.

That said, I can't think of ANY commercial or physical reason to route local Australian traffic through any other country ... especially given the huge fees and charges the US end of the equation adds for traffic. MountainView and other mega-nodes in the US used to be (and probably still are) critical to getting Australian traffic to the world ... but there's no commercial or technical reason I can think of to route purely domestic Australian traffic through them.

I could of course be badly mistaken, but I'm assuming that some of the variables mentioned above apply to Canada (and am sure that they apply in Europe). Especially given their close geographic and network locations to each other. Some of that ex-country traffic may be a simple exercise of the routers determining traffic through an another country's routers would be faster than an alternate domestic route, but a lot of it may be forced by the provider or telco.

In these days of multinational corporates and transnational operations and agreements there are any number of reasons why internal network traffic could cross international borders ... ranging from network exigencies, business priorities, economics, corporate tax evasion ... "Nope, No VAT is due on that purely international transaction", business relationships and arrangements with peers, spying (as Snowden and others have pointed out) or even be routed for more nefarious purposes.

Just my 2 cents worth ...
---
On 29 Apr 2014, at 8:49 am, Roger Clarke <Roger.Clarke at xamax.com.au> wrote:

> A colleague in Canada has conducted an interesting project on:
> 
>     Data Privacy Transparency of Canadian ISPs:
>     http://ixmaps.ca/transparency.php
> 
> Among other things, it co-opts the 'boomerang' concept:  "A boomerang 
> route is a data packet path that starts and ends in Canada, but 
> travels through the USA for part of the journey".
> 
> Angela Merkel's Schengen Net notion addresses the same issue from a 
> European perspective.
> 
> He's asked me about the Australian situation.
> 
> To what extent does traffic from an end-point in Australia to another 
> end-point in Australia travel outside Australia?
> 
> Is that only via the USA, or are there intermediaries in Asia as well?
> 
> And to what extent does traffic from an end-point in Australia to 
> another end-point in Australia travel entirely within Australia but 
> pass through one or more devices controlled by companies that are 
> subject to extra-territorial reach by another government?
> 
> (Naturally there's the USA, with its PATRIOT Act, FISAA and now 
> search-warrant-based demands on all US companies operating 
> everywhere.  But there's also Singtel Optus, and there's Huawei.  And 
> maybe other instances?).
> 
> 
> -- 
> Roger Clarke                                 http://www.rogerclarke.com/
> 
> Xamax Consultancy Pty Ltd      78 Sidaway St, Chapman ACT 2611 AUSTRALIA
> Tel: +61 2 6288 6916                        http://about.me/roger.clarke
> mailto:Roger.Clarke at xamax.com.au                http://www.xamax.com.au/
> 
> Visiting Professor in the Faculty of Law            University of N.S.W.
> Visiting Professor in Computer Science    Australian National University
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link